Shiben Li | Searcharxiv

arXiv2021

A machine learning model to classify dynamic processes in liquid water

The dynamics of water molecules plays a vital role in understanding water. We combined computer simulation and deep learning to study the dynamics of H-bonds between water molecules. Based on ab initio molecular dynamics simulations and a newly defined directed Hydrogen (H-) bond population operator, we studied a typical dynamic process in bulk water: interchange, in which the H-bond donor reverses roles with the acceptor. By designing a recurrent neural network-based model, we have successfully classified the interchange and breakage processes in water. We have found that the ratio between them is approximately 1:4, and it hardly depends on temperatures from 280 to 360 K. This work implies that deep learning has the great potential to help distinguish complex dynamic processes containing H-bonds in other systems.

cond-mat.dis-nn↗

arXiv2021

Neural Network Model for Structure Factor of Polymer Systems

As an important physical quantity to understand the internal structure of polymer chains, the structure factor is being studied both in theory and experiment. Theoretically, the structure factor of Gaussian chains have been solved analytically, but for wormlike chains, numerical approaches are often used, such as Monte Carlo (MC) simulations, solving modified diffusion equation (MDE), etc. In those works, the structure factor needs to be calculated differently for different regions of the wave vector and chain rigidity, and some calculation processes are resource consuming. In this work, by training a deep neural network (NN), we obtained an efficient model to calculate the structure factor of polymer chains, without considering different regions of wavenumber and chain rigidity. Furthermore, based on the trained neural network model, we predicted the contour and Kuhn length of some polymer chains by using scattering experimental data, and we found our model can get pretty reasonable predictions. This work provides a method to obtain structure factor for polymer chains, which is as good as previous, and with a more computationally efficient. Also, it provides a potential way for the experimental researchers to measure the contour and Kuhn length of polymer chains.

cond-mat.soft↗